Comment by jlongr

If the LLM reached 95% accuracy and humans typically hit below 90% accuracy, then it would be demonstrable in court that the accountant who didn't utilize AI or use it as a basis for their marginal enhancements past 95% accuracy was negligent.